Synthesis of the 5/5-spiroindimicin alkaloids: development of a general synthetic approach and biological investigations

Abstract

We describe the development of a unified synthetic strategy for the preparation of all known 5/5-spirocyclic spiroindimicin (SPM) alkaloids, namely spiroindimicins B–G. The present synthetic route relies on four fundamental transformations: Grignard-based fragment coupling between halogenated pyrrolemetal and isatin partners, Suzuki coupling to generate a triaryl scaffold encompassing all requisite skeletal atoms of the natural products, Lewis acid-mediated spirocyclization to construct the 5/5-spirocyclic core, and chemoselective lactam reduction. The developed syntheses are step-economic (6–7 steps from commercial materials), scalable, and amenable to analogue synthesis. Preliminary investigations into a catalytic asymmetric spirocyclization towards an enantioselective SPM synthesis are also described. Further studies of the antiparasitic properties of this class have revealed promising activity against T. brucei for certain congeners. Together with our prior approach to the 6/5-family members, our work constitutes a synthetic solution to all known spiroindimicin natural products.

Graphical Abstract:

Introduction:

In nature, the oxidative dimerization of tryptophan serves as an important branch point to a diverse collection of compound architectures with a wide range of intriguing pharmacological properties.¹ Among this large set of natural products, the spiroindimicins (SPMs, 1–8, Fig. 1A) feature a rare nonplanar, chlorinated heteroaromatic framework incorporating a spirocyclic indoline or indolenine motif bearing a congested quaternary stereocenter connected to pyrrole and indole rings. In 2012, the initial four members of this bisindole alkaloid family, spiroindimicins A–D (1–4), were isolated by Zhang and coworkers from a marine Streptomyces strain.^2a The authors proposed a biosynthetic pathway beginning with lynamicin A (10) or D (11), known achiral dimers of tryptophan (9) co-isolated with 1–4, that could undergo oxidative spirocyclization to generate either a 6/5-spirocyclic skeleton present in SPM A (1) or a 5/5-spirocycle found in the other three congeners (SPMs B–D, 2–4; Fig. 1B). In 2017, the Luzhetskyy group reported^2b the isolation of two monochlorinated members SPMs E (5) and F (6), while Zhang et al. later isolated two deschloro family members, spiroindimicin G (7) and H (8), from a bacterial mutant with an inactivated halogenase gene.^2c The latter investigations, along with earlier studies,³ provided experimental support for the proposed SPM biosynthesis through the expression and validation of several enzymes in the biosynthetic gene cluster. Unfortunately, Zhang et al. were unable to identify the enzyme(s) responsible for the key oxidative spirocyclization.³ In addition to their unique three-dimensional structures, preliminary biological investigations of the SPMs found that they displayed moderate cytotoxicity against several cancer cell lines (IC₅₀ = 9–44 μM).^2a,c

Figure 1.

(A) The spiroindimicin alkaloids. (B) Proposed biosynthetic pathway. (C) Prior total syntheses of spiroindimicins. (D) Our prior total syntheses of 6/5-spiroindimicins.

Based on these preliminary biological activities and their intricate architectures, the spiroindimicins have proven to be attractive synthetic targets.^4,5 So far, three distinct approaches toward the spiroindimicin alkaloids have been successfully demonstrated. In 2015, Sperry and Blair disclosed the inaugural racemic total synthesis of 5/5-spirocyclic spiroindimicins B (2) and C (3) in 15–16 steps from commercially available materials.^4a Their synthetic route relied primarily upon three key steps: an early-stage intramolecular Heck reaction to generate the quaternary spirocenter (13 → 14), a subsequent Fischer indolization to assemble the pentacyclic backbone (14 → 15), and finally a Schöllkopf–Magnus–Barton–Zard (SMBZ) reaction to install the remaining pyrrole ring (17 → 18, Fig. 1C). Later, our group accomplished the first total syntheses of the 6/5-spiroindimicins A (1) and H (8), including an asymmetric synthesis of (+)-1, by employing a fragment coupling approach followed by a carefully choreographed Pd-catalyzed spirocyclization to construct the congested quaternary spirocenter (Fig. 1D).^4b In addition, preliminary biological screening of these SPMs and their unnatural congeners was conducted, revealing their antiparasitic activity against P. falciparum, T. brucei, and L. amazonensis, parasites relevant to the human diseases malaria, African trypanosomiasis, and leishmania, respectively. Recently, Xu and co-workers demonstrated a biomimetic chemoenzymatic approach to racemic 6/5-SPMs A (1) and H (8), and their 5/5-congeners SPMs D (4) and G (7, Fig. 1C).^4c Their concise route is based upon a one-pot dual enzyme-catalyzed oxidative dimerization of L-5-chlorotryptophan (19) or L-tryptophan (9) to afford the symmetrical bisindoles lynamicin D (11) and lycogarubin C (12), which, following indole N-protection (20 and 21), underwent biomimetic spirocyclization under two sets of oxidative conditions to provide the 6/5- and 5/5-SPM scaffolds in a regiodivergent fashion.

Despite their seminal advances, the aforementioned approaches to the 5/5-spiroindimicins present certain limitations. For example, Sperry’s strategy is relatively lengthy and currently only applicable for spiroindimicins bearing a mono(methoxycarbonyl)pyrrole unit.^4a On the other hand, since Xu’s approach is built upon a biomimetic dimerization to symmetrical lynamicin-type intermediates, followed by their oxidative spirocyclization, it will likely be challenging to access the nonsymmetrical spiroindimicins E (5) and F (6) in a selective fashion.^4c Hence, a modular approach applicable to the synthesis of all known 5/5-spiroindimicin family members is desirable yet remains underdeveloped. In continuation of our interest in this family, we disclose herein the full details of the development of our modular synthetic solution to the 5/5-spiroindimicins.⁶ We additionally describe the antiparasitic properties of the 5/5-spiroindimicins for the first time, as well as efforts to render our synthetic approach enantioselective through chiral acid catalysis. Finally, as a demonstration of the robustness of the developed route, we showcase a hundred milligram-scale synthesis of the most active member, spiroindimicin G (7).⁷

Results and discussion:

At the outset of our studies toward the 5/5-spiroindimicins, we initially considered adapting our existing palladium-catalyzed spirocyclization route developed in the context of the 6/5-members (Fig. 1D). However, it quickly became apparent that such an approach might be ‘over-engineered’ in the context of the 5/5-SPMs given their more ‘natural’ indole-indoline connection at their spirocenter, linking the reactive indole C2 and C3 positions (as opposed to C4 and C3 for the 6/5-SPMs) and likely not requiring prefunctionalization at these sites. We therefore formulated a distinct approach built upon different modular fragment couplings between the three component heteroaromatic units, whereafter a spirocyclization leveraging this natural indole reactivity might construct the full 5/5-SPM framework. Through simple variation of these heteroaromatic fragments, we hoped to access any 5/5-member, or designed analogues for structure–activity relationship (SAR) studies.

Retrosynthetically, we envisaged that the monoester SPMs B, C, E, and F (generalized as 26) might arise from the diester SPM D or an appropriate derivative (generalized as 27) via selective demethoxycarbonylation of its C-2 methyl ester in the presence of the more hindered C-5 ester, potentially through a selective monohydrolysis followed by decarboxylation (Scheme 1). The diester SPMs 27 were seen to arise from spirocyclic oxindole 28, incorporating the full 5/5-SPM framework, through chemoselective lactam reduction in the presence of the two methyl esters. The oxindole 28 could be synthesized from 3-hydroxyoxindole 30 via an acid-mediated N-Boc deprotection/Friedel–Crafts spirocyclization reaction forging the crucial 5/5-spirocycle and its highly congested C-3′ quaternary stereocenter. Here, a Brønsted or Lewis acid should promote ionization of the C-3′ tertiary alcohol to generate carbocation equivalent 29 that might be subsequently attacked by the C-2″ center of the pendant indole ring. While such a transformation appears simple on paper and has some related literature precedent,⁸ it is worth noting that this requires the formation of a highly strained 5-membered ring comprised of four sp² carbon atoms. Despite these concerns, the direct nature of this spirocyclization as well as the potential to render the synthesis asymmetric through the use of chiral Brønsted or Lewis acid catalysis⁹ made it an attractive pathway to explore. Precursor 3-hydroxyoxindole 30 encompasses all 3 requisite aromatic fragments present in the target SPMs and was envisioned to arise through two consecutive C–C bond formations. In the forward sense, a functionalized pyrrolemetal species 33 might add to the ketone of isatin 34 to link the first two fragments as adduct 32, followed by chemoselective cross-coupling of its halopyrrole moiety with 3-metallated indole 31 to install the remaining indole.

Scheme 1.

Our modular strategy to access all 5/5-spiroindimicins.

Based on this plan, we commenced our synthetic studies targeting diester SPMs D (4) and G (7, Scheme 2). Commercially available methyl 2-pyrrolecarboxylate (35) was readily converted to iodopyrrole 37 on gram scale by employing reported protocols. This involved initial Fe-catalyzed C–H methoxycarbonylation¹⁰ to the corresponding diester followed by diiodination (NIS, DMF, 80 °C) to diiodopyrrole 36; a controlled monodeiodination of 36 with Zn then delivered 37 in good yield (74%).¹¹ The treatment of iodopyrrole 37 with Knochel’s¹² turbo-Grignard i-PrMgCl•LiCl (2.2 equiv.) at −40 °C generated the corresponding dianionic pyrrolemagnesium species via N–H deprotonation and iodine–magnesium exchange. Addition of the N-methyl 5-chloroisatin¹³ (38a, 1.2 equiv.) electrophile to the reaction mixture effected the desired C–C bond formation, producing the tertiary alcohol 39 in 83% yield. In this case, Knochel’s magnesium–halogen exchange protocol proved superior to lithium–halogen exchange with n-BuLi, presumably due to greater functional group compatibility. Next, a base-mediated iodination (KOH, I₂) of 39 furnished compound 40 (59%) bearing an iodopyrrole moiety suitable for cross-coupling with an appropriate indole unit.

Scheme 2.

(A) Initial approach to diaryl iodide precursor 40 for spiroindimicin D (4). (B) Development of a concise approach to SPM D (4). (C) Extension of the strategy to the deschloro SPM G (7).

While 40 could be accessed in this manner, the above sequence featured an inelegant removal and reinstallation of an iodide handle at the C-3 position of pyrrole ring that lengthened the route. We considered, therefore, whether it might be possible to effect the same deprotonation/iodine–magnesium exchange protocol on diiodide 36 to give a dianion 41 that could add to isatin 38a in a similar way, saving two steps en route to 40. To our delight, when we subjected diiodide 36 to i-PrMgCl•LiCl and 38a under the identical reaction conditions developed for 39, we were able to obtain iodide 40, albeit in a low yield (30%). A brief optimization revealed that the yield improved significantly when we conducted the magnesium–iodine exchange at higher temperature (−10 °C), increased the equivalents of isatin 38a (2.0 equiv.), and azeotropically dried both partners beforehand. Under these conditions, an improved yield of 40 (76%) was achieved, which proved suitable for gram-scale synthesis. It should be noted that no competitive formation of deiodinated product 39 was observed.

With a streamlined approach to access diaryl iodide 40, we next sought to install the remaining indole fragment via cross-coupling. Unfortunately, several standard Suzuki and Stille cross-coupling conditions failed to provide desired C–C coupled product 42, with the diaryl iodide 40 remaining unreacted in most cases. One of the roadblocks to effecting the desired coupling was the poor solubility of iodide 40 in many solvents (e.g., THF, PhMe, 1,4-dioxane) commonly utilized for these cross-coupling reactions (Table1, entries 1 and 2). After screening various solvent systems in the context of a Suzuki coupling with indole boronate 23, DMF with H₂O as a co-solvent (20:1 v/v) was found to be optimal in terms of solubility and reactivity. A broad range of catalyst/ligand systems were then tested (see entry 3 and the box below Table 1), from which Buchwald’s SPhos Pd G4¹⁴ emerged as the optimal precatalyst for this fragment coupling, particularly when paired with K₃PO₄ as base at 40 °C. Application of these conditions afforded the Suzuki coupling product 42 in 52% ¹H NMR yield (entry 5).

Table 1.

Optimization of the assembly of triaryl 42

Entry	ArBPin (equiv)	Pd/L n	Base (equiv)	Solvent	Temp. (°C)	Yield (%)
1	1.8	Pd(PPh3)4	K3PO4 (3.6)	dioxane/H2O	80	NDb
2	1.8	Pd(OAc)2/SPhos	K3PO4 (3.6)	dioxane/H2O	80	NDb
3	1.8	Pd(dppf)Cl2	Cs2CO3 (3.6)	DMF	40	5
4	1.8	SPhos Pd G4	Cs2CO3 (3.6)	DMF/H2O	40	39
5	1.8	SPhos Pd G4	K3PO4 (3.6)	DMF/H2O	40	52
6	1.8	SPhos Pd G4	K3PO4 (3.6)	NMP/H2O	40	NDb
7	1.8	SPhos Pd G4	K3PO4 (3.6)	DMA/H2O	40	28
8	2.0	SPhos Pd G4	K3PO4 (3.6)	DMF/H2O	40	72c
9	2.5	SPhos Pd G4	K3PO4 (3.6)	DMF/H 2 O	40	81 d

^aAll reactions were carried out on 0.035 mmol of 40; ¹H NMR yield with CH₂Br₂ as internal standard.

^bND = not detected.

^cReaction on 0.43 mmol of 40.

^dReaction on 0.82 mmol of 40.

Increasing the amount of the boronate coupling partner 23 slightly (entries 5 and 8) significantly improved the yield, and an optimal 81% yield was achieved with 2.5 equiv. of 23 (entry 9). Since indole boronate 23 suffers from competitive protodeborylation and dimerization during the reaction, we believe the additional 23 helps maintain an effective concentration of this partner despite these deleterious pathways.¹⁵ Interestingly, coupled product 42 exists as a 1:1 mixture of atropisomers, presumably due to the introduction of an axis of chirality through the formation of the new hindered biaryl bond (such atropisomerism was retained in its deBoc congener 43, vide infra).¹⁶

Having assembled all three aryl fragments of SPM D (4), we next explored the acid-mediated Friedel–Crafts-type spirocyclization to provide the full 5/5-spiroindimicin framework in the form of oxindole 44. This process would likely involve initial N-Boc deprotection followed by ionization of the tertiary alcohol to a carbocation equivalent (stabilized by both the oxindole nitrogen and adjacent pyrrole; see 29, Scheme 1) and cyclization of the adjacent indole ring at C-2 to form the key C–C bond. We initially explored treatment of 42 with varying quantities of Brønsted acid (e.g., p-TsOH•H₂O, TfOH, TFA, HClO₄) in chlorinated solvents (DCM, DCE). Unfortunately, these screens were not encouraging as in most cases either 42 remained unreacted or reaction did not proceed beyond N-Boc deprotection (i.e., 43); under more forcing conditions or prolonged times the reaction profiles became complicated with a low yield of 44 obtained in one case (p-TsOH•H₂O). We thus turned our attention to Lewis acid promoters (Table 2). In general, these species were found to be more effective than Brønsted acids, affording the desired spirocycle 44 albeit in low yield (entries 1–7, see Supplementary Material for full details). Temperature played a pivotal role for improving conversion, with higher temperature resulting in a higher yield of 44. A brief solvent screen (DCE, PhCF₃, PhMe, THF, CH₃CN) revealed DCE as the optimal solvent. Overall, BF₃•OEt₂ appeared to be the most effective promoter as it maximized the yield of spirocycle 44 while minimizing the amount of deBoc product 43. Although in principle 43 could lead to more spirocycle 44 over time, this conversion needed to be balanced with competitive decomposition of the product 44 with prolonged reaction times. Ultimately, optimization revealed that subjecting 42 to a substoichiometric amount (0.5 equiv.) of BF₃•OEt₂ in DCE at 70 °C rapidly (20 min) delivered 44 in 58% ¹H NMR yield without any 43 (entry 9). On a preparative scale (60 mg), the use of a higher BF₃•OEt₂ loading (1.0 equiv.) as well as slightly longer reaction time (30 min) proved to be beneficial, affording 44 in 54% isolated yield.

Table 2.

Optimization of the spirocyclization of 42 to SPM D oxindole 44

Entry	Lewis Acid (equiv)	Solvent	Temp. (°C)	Time	43 (%)a	44 (%)a
1	Sc(OTf)3 (1.0)	DCE	70	30 min	11	31
2	ln(OTf)3 (1.0)	DCE	70	1 h	19	22
3	Cu(OTf)2 (1.0)	DCE	70	4 h	–	42
4	TMSOTf (1.0)	DCE	23	1 h	n.d.b	22
5	Zn(OTf)2 (1.0)	DCE	85	24 h	83	11
6	Ce(OTf)3 (1.0)	DCE	85	24 h	55	38
7	BF3•OEt2 (1.0)	DCE	70	30 min	–	43
8	BF3•OEt2 (1.0)	DCE	70	5 min	–	55
9	BF3•OEt2 (0.5)	DCE	70	20 min	–	58
10	BF3•OEt2 (1.0)	PhMe	70	24 min	12	54

^aAll reactions were carried out on 0.02 mmol scale; ¹H NMR yields with CH₂Br₂ as internal standard.

^bNot determined.

With the full 5/5-spiroindimicin framework constructed, the final task that remained to access SPM D (4) was the chemoselective reduction of the oxindole of 44 to an indoline in the presence of the two methyl ester units. Our initial efforts employing borane reduction or Nagashima’s Pt-catalyzed amide reduction protocol were unsuccessful. We ultimately found that conditions reported by Dixon^17b involving Ir-catalyzed hydrosilylation with Vaska’s catalyst and TMDS (originally introduced by Nagashima)^17a were fit for purpose. Here, the reaction proceeds in two phases: an initial Ir-catalyzed oxindole hydrosilylation to generate a sensitive O-silyl hemiaminal intermediate, which could be further reduced by the addition of NaBH₃CN (AcOH/MeOH) to the same pot.^17b This reduction afforded the racemic natural product 4 in 73% yield, completing our total synthesis of spiroindimicin D (4) in 6 steps from commercially available pyrrole 35 (longest linear sequence) with 16.0% overall yield. Notably, our synthetic route appears reasonably scalable, with a single run providing >50 mg of SPM D.

To further validate the modularity of this protocol, we embarked on the total synthesis of closely related deschloro member spiroindimicin G (7). Here, commercially available N-methylisatin (38b) and indole boronate 47 were utilized as coupling partners for the two C–C fragment coupling steps to afford triaryl intermediate 48 in good yield under the previously developed conditions (Scheme 2C). 48 was able to be advanced via the same spirocyclization and reduction transformations, with minor modification to the reaction conditions, providing (±)-SPM G (7) in 6 total steps (15.0% overall yield).

After successful completion of both diester 5/5-spiroindimicins D and G, the stage was set to investigate the synthesis of monoester SPM B (2) from SPM D (4) via regioselective monodemethoxycarbonylation at the C-2 position in presence of the C-5 methyl ester unit (Scheme 3). The selectivity of this approach hinged on the presumed steric effect of the spirocyclic indoline moiety of 4, where this unit and its associated quaternary center might potentially shield the nearby C-5 ester unit from hydrolysis relative to the more accessible C-2 ester. It is necessary to mention here that, prior to these investigations, we conducted 1D NOE experiments of SPM D (4) to validate the isolation group’s assignment of each methyl ester unit, finding that these had been incorrectly assigned. Specifically, we saw a correlation between the aryl C-5″ proton (8.12 ppm) with C-9 methyl ester Me signal (4.09 ppm; originally assigned as C-7), meaning that the original C-9/C-7 OMe assignments should be swapped.^2a This meant that for the desired hydrolytic selectivity to be achieved we would hope to see loss of the C-9 OMe signal, corresponding to hydrolysis of the C-2 methyl ester group, with no NOE correlation to C-5″ from a methyl ester signal remaining.

Scheme 3.

Attempted regioselective ester removal in SPM D (4) towards SPM B (2).

To realize this transformation, SPM D (4) was treated with KOH (1.0–1.2 equiv.) in MeOH/H₂O at 75 °C overnight. Contrary to our hypothesis, hydrolysis occurred preferentially at the undesired C-5 ester, producing 49 as major product (confirmed by the retention of the C-5″ NOE correlation) along with desired product 50 and diacid 51 as minor products. The product ratio (49/50/51) varied from 4:1:0.4 to 4.5:1:1.7 depending on the equivalents of KOH utilized. To explain the observed selectivity, we believe the conjugation with the indole nitrogen reduces the electrophilicity of C-2 methyl ester (see 54) compared to the C-5 ester (which cannot achieve similar extended conjugation). Thus, despite the anticipated steric hindrance of the indoline moiety, preferential hydrolysis occurred at the more electrophilic C-5 ester. Additionally, thermolysis (180 °C)^4b,18 of the crude hydrolysis mixture under neat conditions failed to provide SPM B (2) or even undesired decarboxylated products 52 and 53. Attempts to decarboxylate under other conditions (see Supplementary Material for details) were also unsuccessful.

Given the failure of this regioselective monodemethoxycarbonylation, we planned to amend our approach to the monoester SPMs. Specifically, we sought to leverage the modularity of our fragment coupling strategy and replace the prior diester halopyrrole species 36 with a corresponding monoester variant 56 which might participate in the same key C–C bond formations en route to monoester spiroindimicins B, C, E, and F (Scheme 4A).

Scheme 4.

(A) Modified route to access monoester spiroindimicins. (B) Initial approach toward diaryl bromide precursor 63 for SPM B (2). (C) Total synthesis of SPM B (2).

As our initial monoester fragment 56, we turned to N-TIPS dibromopyrrole ester 59, which was known to undergo selective lithium–bromine exchange proximal to its methyl ester with n-BuLi (Scheme 4B).¹⁹ Dibromide 59 could be accessed from commercial N-TIPS pyrrole (57).²⁰ 57 was first brominated with recrystallized NBS to generate an unstable tribromopyrrole 58 that was taken forward without further purification.^20a It is important to mention that in this step NBS needs to be added portionwise over 48 h while maintaining the reaction temperature at −78 °C (see Supplementary Material for details), otherwise unwanted side reactions diminished the yield of 58 significantly. Subsequent C-2-selective lithium–bromine exchange on 58 followed by the addition of methyl chloroformate gave target monoester dibromide 59 in 57% yield over 2 steps.²⁰ To effect coupling to N-methyl 5-chloroisatin (38a), the reported ester-directed C-3 lithiation was attempted with n-BuLi, followed by addition of 38a. This produced an addition product that could be isolated in pure form after N-TIPS deprotection with TBAF, albeit in poor yield (61, 12% over 2 steps). Aside from the low yield of the coupling product obtained, the spectral properties of 61 deviated significantly from that of analogous adduct 40 in the diester series, which raised questions about its connectivity. These dual concerns led us to consider switching the order of the lithiation/coupling and deprotection steps (Scheme 4C). Pyrrole ester 59 was thus subjected to TIPS deprotection with TBAF providing N–H pyrrole 62 in good yield (80%, not shown). More step-economically, this deprotection step can be performed in the same pot as the prior methoxycarbonylation step by simply adding TBAF to the reaction mixture once 59 had been formed, producing 62 in a comparable overall yield (48% over 2 steps vs. 46% over 3 steps). Next, similar to our prior pyrrole approach in the diester series, 62 was first treated with n-BuLi (2.0 equiv.) to generate its dianion, which subsequently reacted with isatin 38a to afford a different adduct 63 in 84% yield. Encouragingly, the spectral data of 63 more closely resembled those obtained for diester adduct 40. Ultimately, single crystal X-ray analysis of both adducts 61 and 63 confirmed 63 to have the desired connectivity, whereas 61 was shown to be the constitutional isomer resulting from attack of the corresponding 4-lithiopyrrole onto isatin 38a. While it is unclear at this stage why N-TIPS compound 59 resulted in a different regiochemical outcome in its C–C coupling with 38a, Okano has noted that a similar dibromopyrrole ester can lead to different regioselectivity depending on the particular lithiation conditions used.²¹

With an efficient synthesis of 63 established, we could explore the Suzuki coupling reaction with indole boronic acid ester 23. Unlike iodopyrrole 40 in the diester series, bromide 63 was found to be significantly less reactive and rather challenging to couple with 23. A representative subset of our screening efforts is shown in Table 3 (for full details, see Supplementary Material). Under the conditions previously optimized for 40, we found that product 64 was formed in trace yield at best, even at higher temperatures (entry 1). No significant improvement was observed upon testing different coupling partners, including the corresponding 3-indolyl potassium trifluoroborate (65a), boroxine (65b), or stannane (65c). Screening of a broad range of catalyst/ligand systems (see bottom of Table 3) and temperatures failed to provide a better outcome. Ultimately, the incorporation of a stoichiometric amount of a CuCl additive, as initially described by chemists at Merck,²² improved the efficiency of the Suzuki coupling reaction between 63 and boronate 23. A brief screen of alternate Cu(I) additives found CuCl to be optimal. As suggested by the original authors,²² it is likely that Cu(I) assists in relaying the 3-indolyl fragment to Pd through initial B to Cu then Cu to Pd transmetallation (see inset, Table 3), potentially outcompeting unproductive protodeborylation. Interestingly, unlike the prior Suzuki coupling reaction in the diester series, the inclusion of H₂O as a cosolvent was not beneficial in this case (compare entries 5 and 6). After minor adjustments to other parameters (entries 7–11) including increasing the equivalents of boronate partner 23, CuCl, and base, we were able to reproducibly generate coupled product 64 in 44% NMR yield (entry 9). On preparative scale we were able to obtain triaryl 64 in an improved 54% isolated yield under analogous conditions.

Table 3.

Optimization of the construction of triaryl 64

Entry	Partner (equiv)	Catalyst	Additive (equiv)	Base (equiv)	Solvent b	Temp. (°C)	Time (h)	Yield (%)a
1	23 (2.5)	SPhos Pd G4	–	K3PO4 (3.6)	DMF/H2O	70	24	7
2	65a (2.5)	SPhos Pd G4	–	K3PO4 (3.6)	DMF/H2O	70 to 90	36	–
3	65c (2.5)	Pd(PPh3)4	CuTC (1.5)	–	NMP	23 to 80	36	–
4	65b (0.9)	SPhos Pd G4	–	K3PO4 (3.6)	DMF/H2O	70	24	10
5	23 (2.5)	SPhos Pd G4	CuCl (1.0)	K3PO4 (3.6)	DMF/H2O	70	24	11
6	23 (2.5)	SPhos Pd G4	CuCl (1.0)	K3PO4 (3.6)	DMF	70	24	13
7	23 (2.5)	SPhos Pd G4	CuCl (2.5)	K3PO4 (3.6)	DMF	70	24	22
8	65b (0.9)	SPhos Pd G4	CuCl (2.5)	K3PO4 (3.6)	DMF	70	24	11
9	23 (5.0)	SPhos Pd G4	CuCl (5.0)	K 3 PO 4 (5.0)	DMF	70	24	44
10	23 (5.0)	SPhos Pd G4	CuCl (2.5)	K3PO4 (5.0)	DMF	70	24	29
11	23 (2.5)	SPhos Pd G4	CuCl (5.0)	K3PO4 (3.6)	DMF	70	24	27

^aAll reactions were carried out on 0.035 mmol scale; ¹H NMR yield with CH₂Br₂ as internal standard.

^bFor solvent mixtures the ratio was 20:1.

After successfully preparing triaryl 64, we were poised to apply a similar deprotection/spirocyclization approach to reach the monoester 5/5-SPM framework 66. Unfortunately, monoester 64 proved unstable under the previously developed BF₃-mediated conditions, resulting in rapid decomposition potentially due to its relatively more electron-rich pyrrole unit; trials at lower temperatures also failed to provide spirocycle 66. We therefore tested milder Lewis acidic conditions to effect this key C–C bond formation. Ultimately, the use of Ce(OTf)₃ in a combination of HFIP and DCE at 85 °C was identified as optimal, affording spirocycle 66 in moderate yield (47%). Finally, implementation of the same Ir-catalyzed oxindole reduction furnished (±)-spiroindimicin B (2) in 58% yield, completing its total synthesis in 6 steps and 5.9% overall yield.

We next planned to highlight the applicability of our modular approach to the two non-symmetrical monochlorinated members of this family, SPMs E (5) and F (6), targets that may prove challenging to access via Xu’s biomimetic approach.^4c As before, this would simply involve employing the appropriate isatin or indole boronate coupling partners in the developed fragment couplings, followed by the same endgame (Scheme 5). Pleasingly, substituting either the deschloro indole boronate 47 or isatin 38b for their respective 5-chloro congeners resulted in productive C–C couplings, giving access to triaryl fragments 67 and 69 that could be advanced through the same Friedel–Crafts spirocyclization and reduction sequence as for SPM B (2). The first total syntheses of (±)-spiroindimicins E (5) and F (6) were thus accomplished in 6 steps and 3.8 and 5.6% overall yield, respectively. Additionally, the structure of our synthetic SPM F (6) was confirmed through single crystal X-ray analysis.

Scheme 5.

Modular preparation of spiroindimicins E (5), F (6), and C (3).

For spiroindimicin C (3), the only family member containing a free N–H indoline, we initially envisaged direct access from SPM B (2) or its oxindole precursor 66 via N-demethylation. Unfortunately, preliminary investigations along these lines were not fruitful.²³ Thus, once again leveraging the modularity of our route, we sought to incorporate known isatin 38c²⁴ having a removable N-p-methoxybenzyl (PMB) group. Employing 38c in the usual fragment coupling gave adduct 70 in 84% yield. Advancing 70 through the same Suzuki coupling, spirocyclization, and reduction steps afforded N-PMB spiroindimicin C (74, see Table 4). A brief screen of deprotection conditions identified that the PMB group could be removed via hydrogenolysis under acidic conditions²⁵ (H₂, Pd/C, HCl, MeOH) without affecting the aryl chlorides, providing the natural product (±)-SPM C (3) in 46% yield (7 steps and 2.2% overall). The spectroscopic data for our synthetic SPMs B–D (2–4) and G (7) matched well with the data for the natural compounds provided by Zhang.^2a,c In contrast, our NMR data for SPMs E (5) and F (6) differed slightly in some cases from that reported by Luzhetskyy et al. in their reported solvent(s) (data did match in an alternate solvent in some cases), which might be attributable to a typographical error (see Supplementary Material for further details).^2b,26 As noted above, we were able to obtain structural confirmation of our synthetic (±)-spiroindicimin F (6) via X-ray crystallography.

Table 4.

Antiparasitic screening of synthetic spiroindimicins and oxindole congeners^a

	Antiparasitic Activity			Selectivity
Compound	T. brucei EC50 (μM)	P. falciparum 3D7 EC50 (μM)	L. amazonensis EC50 (μM)	HepG2 CC50 (μM)
4: (±)-SPM D	3.4 ± 0.21	5.6 ± 0.22	7.7 ± 0.23	49 ± 1.4
7: (±)-SPM G	0.65 ± 0.14	7.6 ± 0.47	8.1 ± 0.50	47 ± 3.1
2: (±)-SPM B	3.0 ± 0.23	3.0 ± 0.20	8.0 ± 0.49	37 ± 4.3
3: (±)-SPM C	13 ± 0.98	7.1 ± 0.58	>10	>50
5: (±)-SPM E	4.4 ± 0.32	4.8 ± 0.51	9.3 ± 1.7	49 ± 1.0
6: (±)-SPM F	4.0 ± 0.20	9.0 ± 0.97	9.9 ± 1.2	>50
44	4.5 ± 0.73	2.9 ± 0.64	>10	21 ± 1.7
66	11 ± 1.09	11 ± 0.67	>10	47 ± 1.7
71	1.5 ± 0.05	4.1 ± 0.10	6.7 ± 0.62	>50
72	7.6 ± 0.36	7.3 ± 0.52	>10	>50
73	14 ± 2.52	13 ± 0.88	>10	47 ± 3.1
74	1.2 ± 0.03	3.7 ± 0.35	>10	>50
paclitaxelb	0.0076 ± 0.00046	–	–	–
DSM265b	–	0.0084 ± 0.00062	–	–
brefeldin Ab	–	–	–	0.064 ± 0.0045

^aData represent the mean EC₅₀ ± standard error for 3 biological replicates. EC₅₀ calculations for each biological replicate were based on data from technical triplicates.

^bPositive control.

Given that our earlier synthetic studies in this alkaloid family had uncovered promising antiparasitic activity for the 6/5-spiroindimicins and their analogues,^4b we submitted our synthetic 5/5-spiroindimicins as well as their oxindole congeners to similar screening against T. brucei, P. falciparum, and L. amazonensis, causative agents for African sleeping sickness, malaria, and leishmaniasis, respectively (Table 4). Our synthetic SPMs and analogues showed moderate antimalarial activity in these assays, with activities against P. falciparum falling in the EC₅₀ = 2.9–13 μM range. While the 6/5-spiroindimicins had proven most potent against the trypanosomatid parasite L. amazonensis (e.g., SPM A (1): EC₅₀ = 1.3 μM),^4b the 5/5-spirocyclic members demonstrated reduced potency (EC₅₀ = 6.7–9.9 μM). Interestingly, this activity appears to be largely confined to compounds containing basic amines (2, 4, 5–7), with only a single exception (71), suggesting that this motif may be important for efficacy against this parasite.

More promising activity was found for the present compounds against T. brucei, a parasite responsible for the neglected tropical disease human African trypanosomiasis (sleeping sickness), a source of significant disease burden in the developing world.²⁷ Here, a select number of compounds showed activities (EC₅₀) around (or even below) 1 μM with no significant cytotoxicity against HepG2 cells (a measure of selectivity). Namely, spiroindimicin G (7) and the PMB-protected spiroindimicin C (74) both showed high efficacy with minimal cytotoxicity against human HepG2 cells (7: EC₅₀ = 0.65 μM, CC₅₀ = 47 μM; 74: EC₅₀ = 1.2 μM, CC₅₀ = >50 μM). These potencies, it should be noted, are comparable to acoziborole (EC₅₀ = 0.6 μM), a leading Phase III candidate for the treatment of trypanosomiasis.²⁸ Given the ease with which our synthetic platform can access these 5/5-spirocyclic compounds, we envisage being able to rapidly generate analogues of these initial leads to delve more deeply into their structure–activity relationships (SAR) against T. brucei.

Encouraged by the promising antiparasitic activity of SPM G (7) against T. brucei, we sought to demonstrate the scalability of our synthetic route towards this most active member (Scheme 6). Pleasingly, we were able to perform the pyrrolemetal addition and Suzuki reactions on 3.5 g and 1.9 g scale, to give 46 and 48, respectively, and the spirocyclization step to 72 on 600 mg scale, with only minimal drop in yield. Additionally, based on contemporaneous investigations into an asymmetric spirocyclization described below, which highlighted the effectiveness of Brønsted acids in MeCN, we were able to develop a higher yielding spirocyclization protocol based on the use of catalytic triflic acid in MeCN at 60 °C. These conditions delivered 72 in 77% yield on 243 mg scale and could also be applied to the spirocyclization en route to SPM D (4), providing an improved yield of 44 (64% vs. 54%), albeit requiring a higher temperature and extended reaction time in this more deactivated system (Scheme 6, bottom). Despite the final oxindole reduction proving slightly more challenging to scale (44% on 117 mg scale vs. 61% on 44 mg), >120 mg of (±)-SPM G (7) was synthesized in total indicating the practicality of the present approach.

Scheme 6.

Scalable synthesis of spiroindimicin G (7).

Finally, to render the present approach to the 5/5-spiroindimicins enantioselective, we investigated an asymmetric spirocyclization promoted by a chiral acid (Table 5). Based on related literature precedent,⁹ our studies focused mainly on the use of chiral Brønsted acid catalysts, where we sought to promote the spirocyclization of N–H indole 75 formed cleanly via thermolytic removal of the Boc group in triaryl 48 (used either crude or in pure form). Initial screens using several BINOL phosphoric acids (76a–c, entry 1) and imidodiphosphate²⁹ (IDP, 76d–f, entry 2) catalysts at varying temperatures failed to provide the spirocyclized product in most cases, with the only exception being 76g which provided 72 in low yield (14%) and enantioselectivity (−9% ee, entry 3).

Table 5.

Investigations toward an asymmetric spirocyclization

Entry	Catalyst	Solvent	Temp (°C)	Yield (%)a	ee (%)a
1b	76a–c	DCE	23 to 80	NR	NR
2	76d–f	DCE	60 to 80	NR	NR
3	76g	DCE	60 to 80	14	−9
4b	76h	DCE	23 to 60	n.d.	26
5b	76i	DCE	23 to 80	n.d.	12
6b	76j	DCE	23 to 80	n.d.	0
7b	76k	DCE	23 to 60	n.d.	30
8b	76k	PhMe	23 to 60	n.d.	21
9b	76k	DCE	60 to 80	NR	NRc
10	76k	EtOAc	60	16	0
11	76k	CH3CN	60	89	0
12	76k	HFIP	60	23	0
13	76k	PhCF3, PhMe	60	messy	n.d.
14	76k	PhCl	60	25	24
15	76k	MTBE	60	NR	NR
16	76k	CH2Cl2	60	48	3
17	76k	DCE	60	26	34
18	76l	DCE	60	23	23
19	76m	DCE	60	38	21
20	76n	DCE	60	33	8
21	76o	DCE	60 to 80	NR	NR
22	76p	DCE	60	15	34
23	76k	DCE	60	21	32d
24	76k	DCE	60	17	27e

^aAll reactions were carried out on 0.019 mmol scale. NR = no reaction.

^bPerformed on crude 75.

^c4Å MS was used.

^d0.2 equiv catalyst used.

^eH₂O was used as co-solvent.

In comparison to the phosphoric acid and IDP catalysts, the more acidic³⁰ BINOL/SPINOL N-triflyl phosphoramides³¹ 76h–p were more effective catalysts of this spirocyclization reaction, although both yield and enantioselectivity remained moderate. To date, the highest enantioselectivity has been obtained with BINOL N-triflyl phosphoramide (10 mol%) 76k and SPINOL N-triflyl phosphoramide 76p (10 mol%) at 60 °C in DCE (34% ee), in relatively low yield (15–26%). Screening of other solvents using catalyst 76k did not provide better results, although it should be noted that with CH₃CN as solvent 72 was obtained in an excellent 89% yield, albeit in racemic form, which ultimately informed the development of the improved spirocyclization protocol for SPM G (7) outlined above. Screening of a few additives (4 Å MS, H₂O; entries 9 and 24) or chiral Lewis acid complexes (not shown) failed to provide promising results. Future studies will seek to improve upon these initial results.

Conclusions:

In summary, we have developed a unified synthetic approach to prepare all 6 members of 5/5-spirocyclic spiroindimicin alkaloids via highly concise synthetic sequences (6–7 steps) built upon modular fragment couplings.

The brevity and efficiency of our developed route to spiroindimicins B–G (2–7) is the result of significant optimization of each of the three key C–C bond forming events: a functionalized pyrrolemetal 1,2-addition, a hindered Suzuki coupling, and a Friedel–Crafts-type spirocyclization. This carefully choreographed route has proven to be scalable, as demonstrated by the significant quantities of the bioactive (±)-spiroindimicin G (7) prepared to date. Additionally, we have disclosed preliminary investigations to render our approach enantioselective via a chiral Brønsted acid-catalyzed variant of the key spirocyclization, resulting in moderate selectivity that will require further optimization.

Taken together with our prior work towards the 6/5-spiroindimicins, these convergent total syntheses provide a synthetic blueprint to access every member of this intriguing alkaloid class as well as complex analogues. Our synthetic platform has enabled biological testing of the 5/5-spiroindimicins, demonstrating their antiparasitic properties for the first time and offering some preliminary SAR for the family. Their newfound availability, conferred by our synthesis, positions this class well for further biomedical exploration, particularly in the development of novel therapeutics for African trypanosomiasis.

Data availability:

The data supporting this article have been included as part of the Supplementary Material.

Crystallographic data for compounds 46, 61, 63, and 6 have been deposited at the CCDC under accession numbers 2249620, 2255226, 2264967, and 2264968.